Human Peroxiredoxin 1 and 2 Are Not Duplicate Proteins

Human peroxiredoxins 1 and 2, also known as Prx1 and Prx2, are more than 90% homologous in their amino acid sequences. Prx1 and Prx2 are elevated in various cancers and are shown to influence diverse cellular processes. Although their growth regulatory role has traditionally been attributed to the peroxidase activity, the physiological significance of this function is unclear because the proteins are highly susceptible to inactivation by H2O2. A chaperone activity appears to emerge when their peroxidase activity is lost. Structural studies suggest that they may form a homodimer or doughnut-shaped homodecamer. However, little information is available whether human Prx1 and Prx2 are duplicative in structure and function. We noted that Prx1 contains a cysteine (Cys83) at the putative dimer-dimer interface, which is absent in Prx2. We studied the role of Cys83 in regulating the peroxidase and chaperone activities of Prx1, because the redox status of Cys83 might influence the oligomeric structure and consequently the functions of Prx1. We show that Prx1 is more efficient as a molecular chaperone, whereas Prx2 is better suited as a peroxidase enzyme. Substituting Cys83 with Ser83 (Prx1C83S) results in dramatic changes in the structural and functional characteristics of Prx1 in a direction similar to those of Prx2. Here we also report the first crystal structure of human Prx1 and the presence of the Cys83–Cys83 bond at the dimer-dimer interface of decameric Prx1. These findings are consistent with the hypothesis that human Prx1 and Prx2 possess unique functions and regulatory mechanisms and that Cys83 bestows a distinctive identity to Prx1.

Structural studies from various species suggest that Prx proteins may exist as an ␣ 2 homodimer or a doughnut-shaped (␣ 2 ) 5 homodecamer (23)(24)(25). Two molecules of reduced Prxs form an ␣ 2 homodimer, which in turn serves as a building block for decamer formation. According to x-ray crystallographic studies, the inter-molecular disulfide bond that occurs transiently during the catalytic cycle requires local unfolding because the distance between the catalytic Cys and the resolving Cys is too far apart at ϳ13 Å (24,25). The latter effect destabilizes the dimer-dimer (DD) interface and, by extension, the decameric structure. The stability of the DD interface is quickly restored when the disulfide bond is reduced back to Cys-SH by thioredoxin. This completes a cycle of H 2 O 2 catalysis by the Prx protein. If the mechanics of unfolding or the conditions for disulfide formation are unfavorable for various reasons, the catalytic Cys becomes overoxidized to sulfenic (-SO 2 H) or sulfonic acid (-SO 3 H) and loses its H 2 O 2 catalytic activity. The DD interface of an overoxidized Prx dimer, however, appears to be more stable than that of a fully reduced Prx dimer, and it may in fact lead to a more compact decamer (23)(24)(25). Recent studies suggest that high molecular weight Prx may function as a molecular chaperone (26 -28). The information above suggests that the redox status of the catalytic Cys and the molecular chaperone activities of Prx may be closely linked.
Human Prx1 and Prx2 are 91% homologous and 78% identical in their amino acid sequences. Are Prx1 and Prx2 duplicate proteins or do they have unique functions and regulatory mechanisms? Although Prx1 and Prx2 have been studied independently in a number of cell and animal systems, little information is available regarding this fundamental question. Studies of Prx1 knock-out (KO) and Prx2KO mice illustrate the complexity of addressing this deceivingly simple question. One of the major phenotypes of Prx1KO was uncontrolled cell proliferation and the development of tumors in certain cell types (29). This striking observation led to the suggestion that Prx1 may act as a tumor suppressor. This is contradictory to the proposed role of Prx1 in promoting cell survival, treatment resistance, and malignant progression of various cancer cells (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16). Absence of Prx2, on the other hand, resulted in splenomegaly caused by the congestion of red pulp with hemosiderin accumulation in Prx2KO mice (30). No tumor development was observed in any cell types or tissues of Prx2KO mice. Based on this finding, it was suggested that Prx2 is essential for sustaining the life span of erythrocytes in mice. Although there are no data available comparing the levels of H 2 O 2 in different tissues of the Prx1KO or Prx2KO mice, or between the same tissue or cells obtained from Prx1KO and Prx2KO mice, elevated levels of H 2 O 2 were observed in certain tissues of both Prx1KO and Prx2KO mice. Obviously, reduced peroxide removal resulting from the absence of either Prx1 or Prx2 cannot offer an answer to the drastically different phenotypes observed in these studies. The above information and findings from various human cancer cells and tissues, however, clearly indicate the following: 1) Prx1 and Prx2 may not be identical in function and are distinguishable by their regulatory mechanisms; 2) maintaining an adequate level of these proteins may be important in keeping normal cell proliferation/apoptosis in check; 3) the functions of Prx1 and Prx2 may be regulated in a cell type-and tissue contextdependent manner; and 4) their functions would need to be tested (and interpreted) in light of the various genetic and environmental factors that could affect the molecular behavior of Prx1 and Prx2.
When we examined the amino acid sequences of human Prx1 and Prx2, we noted that in addition to the well characterized catalytic Cys (Cys 52 in Prx1 and Cys 51 in Prx2) and resolving Cys (Cys 173 in Prx1 and Cys 172 in Prx2) residues, there is a Cys 83 that is unique to Prx1 and not found in Prx2 (Fig. 1). According to available structural information from various Prx species (24,25), the Cys 83 of human Prx1 appears to be located at the putative DD interface. The goal of this study was to investigate whether there is a functional and/or structural difference between human Prx1 and Prx2 at the protein level, and if so, whether Cys 83 in Prx1 contributes to the difference. Our approach was to employ defined experimental systems and to compare the molecular behaviors of these two proteins side by side. We purified recombinant human Prx1 and Prx2 proteins and evaluated their peroxidase and molecular chaperone activities. We also examined the biochemical and structural characteristics of the Cys 83 3 Ser substituted Prx1 (Prx1C83S) with those of wild type Prx1 and Prx2. Based on the results from a combination of mutagenesis, biochemical, computer modeling, mass spectrometry, and x-ray crystallographic studies, we conclude that Prx1 and Prx2 are not redundant proteins and that the Cys 83 of Prx1 plays a critical role in bestowing a distinctive identity on Prx1, which is not shared by Prx2.

EXPERIMENTAL PROCEDURES
Materials-Dithiothreitol (DTT), iodoacetamide (IAA), NADPH, hydrogen peroxide, citrate synthase, and insulin were obtained from Sigma. Tris-(2-carboxyethyl)phosphine hydrochloride was from Invitrogen. Malate dehydrogenase (MDH) was from Roche Applied Science.  (53). Note that the amino acid corresponding to position 24 of Prx1 is missing in Prx2. The locations of the Cys residues are marked. In addition to the catalytic Cys and the resolving Cys, one extra Cys is shared by Prx1 and Prx2 at positions 71 and 70, respectively. The presence of Cys 83 is unique to Prx1. Identical residues (*) and strong (:) or weak (.) homologous residues are indicated below the sequences. The amino acid residues located at the putative DD interface are underlined.

Expression and Purification of Recombinant Proteins-The
Prx1/pET-17b and Prx2/pET-17b vectors containing the entire coding regions of human Prx1 and Prx2 were constructed. Sitedirected mutagenesis of Cys 83 to Ser 83 was performed to generate the Prx1C83S with the QuikChange mutagenesis kit (Stratagene) using the Prx1/pET-17b as a template. Recombinant human Prx1, Prx2, and Prx1C83S proteins were purified by sequential ion exchange chromatography and size exclusion chromatography as described previously (31). Briefly, the cell extract was loaded onto DEAE-Sepharose (GE Healthcare) and equilibrated with 20 mM Tris-Cl (pH 7.5). The proteins were dialyzed with 50 mM sodium phosphate buffer (pH 6.5). The dialyzed proteins were loaded onto SP-Sepharose (GE Healthcare) and equilibrated with 50 mM sodium phosphate buffer (pH 6.5). The bound proteins were eluted with a linear gradient of sodium chloride. The fractions containing Prx1, Prx2, or Prx1C83S were pooled, loaded onto Superdex 200 (16:60, GE Healthcare), and equilibrated with 50 mM sodium phosphate buffer (pH 7.0) containing 0.1 M NaCl. The fractions containing Prx1, Prx2, or Prx1C83S were pooled and stored at Ϫ80°C. Unless otherwise specified, all proteins were treated with 0.5 mM DTT for 30 min prior to their use. DTT was removed with Hitrap desalting column (5 ml; GE Healthcare). Full-length clones of yeast thioredoxin (yTrx) and yeast thioredoxin reductase (yTR) were obtained from yeast genomic DNA (Clontech). The yTrx and yTR proteins were purified as described previously (32).
Peroxidase Activity Assay-The thioredoxin-dependent peroxidase activity of the purified Prxs was measured as described previously with minor modifications (32). A total of 1.1 g each of Prx1, Prx2, or Prx1C83S was incubated in 50 mM Hepes (pH 7.0) containing 200 M NADPH, 3 M yTrx, and 1.5 M yTR. The reaction mixture was incubated at 30°C for 5 min, followed by the addition of a 10-l aliquot of H 2 O 2 at various concentrations. NADPH oxidation was monitored for the next 10 min by a decrease in absorbance at 340 nm measured with the Ultrospec 2100 pro (GE Healthcare) spectrophotometer.
Molecular Chaperone Activity Assay-Molecular chaperone activity was determined as described previously (33) by assessing the ability of the recombinant Prxs to inhibit the thermal aggregation of substrate proteins (26 -28). Briefly, 1 M of MDH, citrate synthase, or insulin was mixed with various concentrations of Prx1, Prx2, or Prx1C83S, in a degassed 50 mM Hepes (pH 7.0) solution containing 0.1 M NaCl. The reaction mixture was incubated at 45°C for 30 min, and the increase of light scattering as a result of thermal aggregation of substrate proteins was monitored at 360 nm.
Statistical Analysis-Statistical significance was examined using Student's t tests. The two-sample t test was used for two group comparisons. Values were reported as means Ϯ S.D. p values Ͻ0.05 were considered significant.
Western Blot Analysis-A total of 1.1 g each of Prx1, Prx2, or Prx1C83S was incubated in 50 mM Hepes (pH 7.0) containing 200 M NADPH, 3 M yTrx, and 1.5 M yTR. The reaction mixture was incubated at 30°C for 5 min, and various concentrations of 10 l H 2 O 2 were added to the mixture. After 5 min of incubation, 0.6 mol of IAA was added to block the residual free thiol (ϪSH) group. The reaction was allowed to go in the dark for 5 min and then quenched by adding 100 nmol of Tris-(2-carboxyethyl)phosphine hydrochloride. An aliquot of 100 ng of Prxs from each reaction mixture was taken and subjected to Western blot analysis with Prx1, Prx2, or Prx-SO 2/3 H antibodies (Lab Frontier Life Science Institute, Seoul, Korea).
Intrinsic Tryptophan Fluorescence Spectroscopy-Intrinsic Trp fluorescence spectra were recorded using an LB-45 spectrofluorometer (PerkinElmer Life Sciences) with an excitation wavelength of 295 nm. Samples of Prx1, Prx1C83S, or Prx2 protein were used at a concentration of 10 g/ml in 50 mM sodium phosphate buffer (pH 7.0) containing 0.1 mM NaCl. The excitation and emission band passes were set at 3 nm. Spectra were monitored from 300 to 500 nm at room temperature.
Homology Modeling-A model of decameric human Prx1 was built based on the atomic structure of well characterized bacterial Prx protein, AhpC (Protein Data Bank code 1YEP). A molecular operating environment (version 2005.06, Chemical Computing Group, Montreal, Canada) running on a G5 dual 2.7-GHz PowerPC workstation was used for homology modeling. The best intermediate model was energy-minimized to remove unfavorable van der Waals contacts. PyMOL (34) was used for analysis and illustration purposes.
Intact Protein Analysis by MALDI-TOF Mass Spectrometry-Samples of Prx1, Prx1C83S, or Prx2 in water, 0.1% trifluoroacetic acid were mixed with a solution of 3,5-dimethoxy-4-hydroxycinnamic acid, sinapinic acid matrix. Five pmol each of the proteins was applied to a sample plate. The MALDI micro-MX TM mass spectrometer (Waters, Milford, MA) was operated in positive linear mode (post acceleration dynode, 5 kV; sample period, 1 ns). Data acquisition was performed over the m/z ranges of 25,000 -250,000. Mass spectra were calibrated by the default method and verified by using bovine serum albumin for external calibration.
In-solution Digestion of Recombinant Proteins with Trypsin-Samples of Prx1, Prx1C83S, or Prx2 proteins were diluted to 1 M concentration with 10% acetonitrile, 40 mM ammonium bicarbonate buffer. After brief sonication to dissolve the samples, 200 l each of protein solution was transferred to Eppendorf tubes. For some experiments, the protein solution was reduced with 20 l of DTT (50 mM in 10% acetonitrile, 40 mM ammonium bicarbonate buffer) at 60°C for 30 min. After it was cooled to room temperature, iodoacetamide solution was added (20 l, 200 mM in 10% acetonitrile, 40 mM ammonium bicarbonate buffer), and the reaction mixture was incubated at room temperature for 30 min in the dark. A solution of sequencing grade trypsin (10 g of trypsin dissolved in 500 l of 10% acetonitrile, 40 mM ammonium bicarbonate buffer) was added to the mixture to achieve an enzyme/protein ratio of 1:10 (w/w). The solution was digested at 37°C for 16 h.
MALDI-TOF Analysis of Trypsin-digested Peptides-After trypsin digestion, the peptide mixtures were dried in a Speedvac Structural and Functional Differences of Prx1 and Prx2 JULY 27, 2007 • VOLUME 282 • NUMBER 30 concentrator and dissolved in water containing 0.1% trifluoroacetic acid. Samples were mixed 1:1 (v/v) with 10 mg/ml ␣-cyano-4-hydroxycinnamic acid in 400:400:200 (v/v) acetonitrile/ethanol/water. A total of 1.3 l of sample/matrix mixture was spotted onto a MALDI target plate, allowed to dry in air at room temperature, and analyzed on a MALDI micro MX TM mass spectrometer. The machine was operated in a positive reflectron mode with an accelerating voltage of 25 kV, and a time lag focusing delay of 500 ns. Data acquisition was performed over the m/z ranges of 900 -4000. Calibration mixture containing polyethylene glycol in a range of molecular weights and sodium iodide was used to generate a multipoint external calibration. The MS data were externally corrected by using the adrenocorticotropic hormone as a lock mass.
Q-TOF Analysis of Trypsin-digested Peptides-A hybrid quadruple/orthogonal time-of-flight mass spectrometer Q-TOF API US (Waters, Milford, MA) was used to carry out a tandem mass spectrometry (MS/MS) analysis. Two l of digested peptide solution were injected into the LC system (nanoACQUITY UPLC, Waters) that was interfaced with a Q-TOF mass spectrometer at a flow rate of 4 l/min. The solution was pre-concentrated on a symmetry C18 column (180 m ϫ 20 mm; Waters). The peptides were eluted onto a nano-ACQUITY C18 column (100 m ϫ 100 mm; Waters) and were separated over a 75-min period at a flow rate of 400 nl/min with a gradient of acetonitrile solution containing 0.1% formic acid. The samples were directed to the electrospray source. The Q-TOF mass spectrometer was operated in a positive ion mode with a source temperature of 120°C and a cone voltage of 45 eV in V-mode. The collision energies were set at 10 and 30 V for MS and MS/MS scans, respectively. The MS/MS spectra were obtained in a data-dependent acquisition mode.
Crystallization, Data Collection, and Structure Solution-Human Prx1 was crystallized as described previously with minor modifications (35). The crystals were obtained by hanging drop vapor diffusion experiments using a reservoir solution of 100 mM citrate (pH 4.6) and 10% polyethylene glycol at a protein concentration of 15 mg/ml. X-ray diffraction data collection was carried out at the Cornell High Energy Synchrotron Source A-1 beam line. One hundred eighty 1°o scillation frames of data from one crystal were recorded on a CCD detector and processed with HKL2000 and SCALEPACK software packages (36). The space group is P2 1 and the unit cell dimensions are a ϭ 89.3Å, b ϭ 111.1Å, c ϭ 120.4Å, ␤ ϭ 111.8°, with one complete decamer in the asymmetric unit. A total of 94,269 diffraction intensities was measured yielding 33,727 unique data to a maximum of 3.20 Å resolution. The overall R-merge for the data set was 0.152 and was 95.1% complete. The I/I value for the data set was 8.16 overall and 1.75 in the highest resolution shell (3.31-3.20 Å). The structure was determined by the molecular replacement method using the AhpC structure as the search model and refined with Refmac5 in the CCP4 crystallographic package (37). After one cycle of refinement with tight noncrystallographic restraints (1980 amino acids and 15,520 atoms; no solvent molecules were included), the crystallographic R-factor and R-free values were 0.299 and 0.328, respectively. The root mean squared deviations from ideality of bond distances and angles are 0.023 Å and 2.14°, respectively, without any major violation in the Ramachandran plot.

RESULTS
Comparison of Peroxidase and Molecular Chaperone Activities of Prx1 and Prx2-To compare the H 2 O 2 catalytic activities of human Prx1 and Prx2, the standard peroxidase reaction was carried out by using the yeast Trx system (yTrx, yTR, and NADPH) as an electron donor. The activities were evaluated by monitoring the decrease in absorbance at 340 nm (A 340 ) because of the oxidation of NADPH. As shown in Fig. 2, A (Fig. 2C).
Next, we compared the molecular chaperone activities of Prx1 and Prx2 by using MDH as a substrate. As shown in Fig.  3A, Prx1 displayed a robust capacity to suppress the thermal aggregation of MDH. The molecular chaperone activity of Prx1 increased in a concentration-dependent manner, peaking at 12 M. In contrast, the chaperone activity of Prx2 was considerably less than that of Prx1 on an equimolar basis (Fig. 3B). Similar results were obtained when citrate synthase or insulin was used as a substrate (data not shown).  4B). Prx1C83S also appeared to have completely lost its ability to suppress the thermal aggregation of MDH (Fig. 4C). The chaperone activities of wild type Prx1 at 8 and 12 M are shown for comparison.
The Oligomeric Status of Prx1 Differs Significantly from That of Prx2 and Prx1C83S-The results above suggest that the molecular properties of Prx1C83S are significantly changed from those of the wild type Prx1, and in a direction that resembles the characteristics of Prx2. Because the structure of Prxs may be closely linked to the peroxidase and molecular chaperone activities, we next carried out gel filtration chromatography to examine the oligomeric status of Prx1, Prx1C83S, and Prx2. The elution profile of Prx1C83S and Prx2 was significantly different from that of Prx1 (Fig. 5A). The position of Prx1 at its peak corresponded approximately to a molecular mass of 340 kDa, which is somewhat greater than the expected molecular mass (220 kDa) of a decameric structure. The position of Prx1C83S was similar to that of Prx2. The apparent molecular masses of Prx1C83S and Prx2 at their peaks were 57 and 67 kDa, respectively. These values were also slightly higher than the expected molecular mass of a dimer (44 kDa). Considering that hydrodynamic radius influences the elution pattern of a protein on gel filtration chromatography, it is likely that the differences between the expected and observed molecular masses resulted from the differential compactness of these proteins in solution. Nonetheless, the results strongly suggested the presence of differences in the oligomerization characteristics of Prx1 compared with that of Prx1C83S or Prx2.
We also examined the intrinsic tryptophan fluorescence profiles of Prx1, Prx1C83S, and Prx2 (Fig. 5B). The intensities of the Trp fluorescence spectra of Prx1 were much lower than those of Prx1C83S and Prx2. Lower intensities in the Prx1 Trp spectra are indicative of a greater propensity to form an oligomeric structure (38). Consistent with the result obtained from the gel filtration study, the Trp fluorescence spectra indicated that the tertiary packing property of Prx1 is significantly different from that of Prx1C83S or Prx2.

Molecular Mass Determination by Mass Spectrometry Reveals That Prx1 Is Present as a Decamer, whereas Prx2 and Prx1C83S Are
Primarily in a Dimeric Form-MALDI-TOF mass spectrometry was employed to determine the molecular masses of the intact Prx1, Prx2C83S, and Prx2 proteins in high resolution. The molecular mass/charge (m/z) values of 219,746.17, 110,201.47, 73,554.93, and 55,373.06 represent the singly, doubly, triply, and quadruply charged Prx1 decamer, respectively, further corroborating that Prx1 is present predominantly as a decamer (Fig. 6A). The minor peaks at m/z of 44,202.26 and 882,902.29 represent the singly charged Prx1 dimer and tetramer ion, respectively. We found that the decamer signature of Prx1 is completely lost in Cys 83 -to Ser 83 -substituted Prx1. As shown in Fig. 6B, Prx1C83S is present predominantly in a dimeric form. The intense signal shown at m/z value of 44,000.22 represents a Prx1C83S dimer with a charge state of ϩ1. The weak signal of the Prx1C83S tetramer ion is detected at m/z value of 88,286.76. Consistent with the functional similarity between Prx1C83S and Prx2, Prx2 is also present primarily as a dimer (Fig. 6C). The intense peak at m/z of 43,627.27 is a Prx2 dimer with a charge state of ϩ1. Similar to Prx1C83S, a weak signal of the Prx2 tetramer ion (m/z 87,373.28) was also detected. The molecular mass determination of intact protein by MALDI-TOF mass spectrometry is highly accurate. The % errors of all the experimental m/z values were less than 0.5% of the theoretical values (Table 1), except for the singly charged Prx1 decamer ion in which the % error of experimental mass is 0.55%.
The Cys 83 of Prx1 Is Located at the Dimer-Dimer Interface-To gain insight as to how Cys 83 may impact on the structural and functional characteristics of Prx1, we carried out a homology modeling study. Because the crystal structure of human Prx1 has not been determined, the atomic structure of the well characterized bacterial Prx, AhpC (Protein Data Bank code 1YEP), was used as a basis for homology modeling. Our modeling results indicated that the  DD interface of human Prx1 decamer contains a local 2-fold symmetry axis close to Cys 83 of each peptide chain. The Cys 83 residue in either 1 ϭ Ϫ180°or 1 ϭ Ϫ60°conformation packs at the DD interface in a manner similar to the packing of AhpC. The distance between the two Cys residues is at (Cys 83 at 1 ϭ Ϫ180°) or slightly above (Cys 83 at 1 ϭ Ϫ60°) the sum of van der Waals radii, thus ensuring a tight fit and perhaps entropic gain because of the burial of hydrophobic surfaces. Fig. 7 illustrates the DD interface of the Prx1 decamer model with the Cys 83 side chain in 1 ϭ Ϫ180°c onformation. The presence of the Asp 79 and Asp 47 pairs at the DD interface does not appear to be destabilizing, because the side chains are likely be shielded by water molecules at the interface as is the case with the AhpC crystal structure. Consequently, repulsive negative charges at the DD interface or energy costs for burial of charged side chains are probably of little concern.
The Cys 83 Residue of Prx1 Is Oxidized under an Ambient Atmosphere-Because of the specific location of Cys 83 at the DD interface, we questioned whether the disulfide bond between the dimer units contributes to the preferential decamer structure of Prx1. Although the distance between the two dimer Cys 83 sulfur atoms (2.42 Å) is slightly longer than the ideal disulfide bond distance of 2.03 Å (39), a Cys 83 -Cys 83 disulfide bond appears to be plausible with some adjustment to the backbone conformation in this region. We looked for a Cys 83 -Cys 83 -containing peptide ion or the MS/MS fragments of such ion by using MALDI-TOF and Q-TOF mass spectrometry, respectively. We did not find corresponding molecular masses under the experimental conditions that we tested. Nonetheless, when we analyzed the tryptic digests of Prx1 by MALDI-TOF, the 68 -92 peptide that contains the reduced Cys 83 was not detected, indicating that Cys 83 may be oxidized under an ambient atmosphere. As shown in
To test whether the disappearance of the Cys 83 -containing peptide in Prx1 is because of the oxidation of Cys 83 , an aliquot of Prx1 was incubated with DTT, and the thiol (-SH) moieties were alkylated with IAA. The MALDI-TOF analysis revealed that the 68 -92 residue with the m/z of 2881.6497 was fully restored in the DTT-treated and -alkylated wild type Prx1 digests. Fig. 8C shows an overlay of the isotopic clusters of the alkylated 68 -92 peptide ion of Prx1 with or without DTT treatment. In contrast, when the 68 -92 peptide ion of Prx1C83S was alkylated with or without DTT treatment, no difference was observed. Fig. 8D displays an overlay of the isotopic clusters of the alkylated 68 -92 of Prx1C83S at m/z values of 2808.6533 (with DTT) and 2808.5066 (without DTT). These results further confirmed that the oxidation of Cys 83 contributes to the loss of the 68 -92 ion signal of wild type Prx1.
The Cys 83 -Cys 83 Disulfide Bond Is Present at the Dimer-Dimer Interface of Prx1 Decamer-X-ray crystallographic studies were conducted to determine the three-dimensional structure of human Prx1 protein. The crystal structure determined at 3.2 Å (R-factor value of 0.299 and R-free value of 0.328) clearly demonstrated that human Prx1 is present as a decamer, in which five homodimers associate around a local 5-fold rotational axis forming a doughnut-shaped oligomer (Fig. 9A). The crystallographic data also provided direct evidence that the Cys 83 residues at the DD interfaces are all involved in the formation of disulfide bridges across dimers. The electron density maps of the five interfaces show the presence of disulfide bonds at DD interfaces (Fig. 9B), demonstrating that Prx1 is locked into a decameric form through interfacial disulfide bond formation. One homodimeric Prx1 molecule is shown in Fig. 9C, pointing out the locations of interfacial Cys 83 -Cys 83 disulfide bonds. The locations of the Cys 52 and Cys 172 residues are shown for reference. Given that the Cys 52 residue is approximately 13 Å apart from the Cys 173 in each homodimer unit, it is unlikely that the Cys 52 and Cys 172 are linked by a disulfide bond. The overoxidation status of Cys 52 could not be ascertained from the crystallographic information.

DISCUSSION
Human Prx1 and Prx2 have been studied independently in a number of cell and animal systems and have been shown to influence the survival, proliferation, and treatment response of cancer cells. In this study, we demonstrated that human Prx1 and Prx2 possess unique functions and regulatory mechanisms. Prx1 displays a greater molecular chaperone activity than Prx2 as assessed by its ability to inhibit the thermal aggregation of substrate proteins such as MDH. The peroxidase activity of Prx1, on the other hand, is more susceptible to inactivation by H 2 O 2 than that of Prx2. Consistent with this finding, the catalytic Cys residue of Prx1 is more prone to overoxidation than that of Prx2. Amino acid sequence analysis indicates that there is a cysteine residue at position 83 that is unique to Prx1. Our mutagenesis, biochemical, mass spectrometry, computer modeling, and x-ray crystallographic results are congruent with the hypothesis that the redox-sensitive Cys 83 of human Prx1 plays an important role in modifying the molecular characteristics of Prx1. The substitution of Cys 83 to Ser 83 (Prx1C83S) results in dramatic changes in the structural and functional properties of human Prx1 in a direction that aligns more closely to human Prx2.
Our results show that compared with Prx1, the peroxidase activity of Prx2 is not as easily inactivated as Prx1 during a catalytic cycle. Although our studies are done using purified proteins in cell-free systems, we propose that Prx2 may be physiologically better suited as a peroxidase enzyme. This hypothesis is consistent with the recent study demonstrating the H 2 O 2 catalytic activity of Prx2 in regulating PDGF signaling. Prx2, but not Prx1, translocates to the membrane and eliminates H 2 O 2 that was generated upon PDGF receptor activation in response to PDGF binding to the receptor (40). The catalytic Cys residue of Prx2 is not overoxidized in this condition. According to the available x-ray crystallographic information and the results presented in this study, the regional structure surrounding the active site Cys appears to differ between Prx1 and Prx2; the catalytic Cys in Prx1 appears to be surrounded by several hydrophobic residues (41), whereas access to the active site Cys in Prx2 appears to be restricted by Phe 81 of the adjacent Prx2 dimer (23). This may account in part for the resistance of Prx2 to overoxidation during the catalytic cycle. Because the de novo synthesis of H 2 O 2 during PDGF signaling is likely occur in a very restricted region localized around the PDGF receptor, a preferential translocation of Prx2 from the cytosol to the membrane may also contribute to the peroxidase function of Prx2 during PDGF signaling.
The Cys 83 residue of Prx1 is located at the DD interface. Sequence alignment indicates that Cys 83 is conserved only in mammalian Prx1, including human, bovine, rat, and mouse, but  not in bacteria, yeast, or parasite orthologs of Prxs (Fig. 10). We propose that Cys 83 may have been gained later in evolution, perhaps for the functional specialization of Prx1 in mammalian species. Our results indicate that Cys 83 may increase the affinity between the dimers at the DD interface, thereby enabling the formation of a decamer as a preferred structure. How might the substitution of Cys 83 to Ser 83 decrease the stability of the DD interface? A disulfide bridge across the DD interface may be a possibility. Our crystallographic analysis of the molecular struc-ture of human Prx1 demonstrated the presence of Cys 83 -Cys 83 -linked disulfide bond at each of the five DD interfaces of the decameric Prx1. Although our mass spectrometry studies provided an accurate molecular mass corresponding to the Prx1 decamer, we were not able to detect the Cys 83 -Cys 83 -linked peptide ion under the mass spectrometry conditions that were tested. It is possible that our inability to obtain a Cys 83 -Cys 83 peptide ion or its fragments could result from deformation of the peptide during the gas phase ionization by some unknown mechanisms. Because the nonbonded contact distance is very close (the nonbonded S-S distance is Ϸ2.42Å at 1 ϭ Ϫ180°o r Ϸ4.0Å at 1 ϭ Ϫ60°) with a possible entropic gain of a hydrophobic surface, we cannot exclude the possibility that Prx1 could pack tightly at the DD interface without having to form a Cys 83 -Cys 83 covalent linkage across the dimers. A Ser 83 substitution is likely to alter the DD interface in at least two ways. First, it makes the surface less hydrophobic and more polar. Second, the van der Waals radius of oxygen (in Ser 83 ) is 1.40 Å, as opposed to 1.85 Å for sulfur (in Cys 83 ). Thus, Ser will render the DD packing less tight. These consequences would have a destabilizing effect in Ser 83 -substituted Prx1 and make decamer formation energetically less favorable. How might the structural property of Prx1 explain its molecular chaperone activity and greater sensitivity to inactivation by H 2 O 2 ? According to the previously proposed mechanism (24,25), the H 2 O 2 catalytic cycle of Prxs requires a decamer to dimer transition through local unfolding of the loops containing the catalytic Cys 52 and the Cys 173 . If Cys 83 stabilizes the DD interface of Prx1 and discourages local unfolding, the active site Cys engaged in a catalytic cycle would not be able to form an inter-molecular disulfide bond and its subsequent reduction by  thioredoxin to complete the cycle. As a result, Prx1 is more prone to overoxidation, which in turn may lead to a more stable and compact oligomeric structure. This property may explain why Prx1 is a better molecular chaperone than Prx2. An interesting analogy appears to exist in the isoforms of glutathione peroxidase (Gpx). Although several Gpx isoforms act primarily as an antioxidant peroxidase, the phospholipid hydroperoxide Gpx (PHGpx) forms a capsule around sperm mitochondrion in the testis (42,43). The PHGpx in the capsule is oxidatively cross-linked and enzymatically inactive. Because both Prx and Gpx systems utilize NADPH as an ultimate reducing source, the functions of the two peroxidase systems can be complementary or inter-dependent in certain cell types and tissues. Similar to the Prx system, functional specialization of the Gpx system might have been developed during evolution. Lines of evidence suggests that overoxidation of the catalytic Cys may allow a mechanism of structural and functional switching of Prx from a peroxidase enzyme to a molecular chaperone. This hypothesis is consistent with the behavior of Prx1 in interacting physically with various cellular proteins. Using a yeast two-hybrid system, the interaction of Prx1 with the Src homology 3 domain of c-Abl, the Myc Box II domain of c-Myc, and the macrophage inhibiting factor has been demonstrated (44 -46). We recently reported that Prx1 suppresses radiation-induced JNK signaling and apoptosis in lung cancer cells (31). Our results demonstrated that the JNK inhibitory effect is mediated through the interaction of Prx1 with the GSTpi-JNK complex, thereby preventing JNK release from the complex. The interaction of Prx1 with growth regulatory and signaling proteins may be responsible for the wide range of effects attributed to Prx1 (29,(47)(48)(49)(50). Whether Prx2 behaves similarly to Prx1 in this respect is presently unknown.
In addition, the regulation of Prx1 and Prx2 expression appears to vary widely among different cell types. In the brain, Prx1 is expressed in astrocytes, whereas Prx2 is expressed in neurons (51). Prx1 is preferentially expressed in the Leydig cells, whereas Prx2 predominates in the Sertoli cells in the testis (52). These observations indicate that the cell type-and tissue-specific expression of Prx1 and Prx2 would also contribute to their respective activities in certain cells and tissues, but not in others. In this study, we provide FIGURE 10. Multiple sequence alignments of mammalian Prx1, Prx2, and Prx orthologs across species. The respective Prx1 and Prx2 amino acid sequences of human (hPrx1 and hPrx2), rat (rPrx1 and rPrx2), mouse (mPrx1 and mPrx2), and bovine (bPrx1 and bPrx2) are aligned with those of Saccharomyces cerevisiae (yTSA1 and yTSA2), Crithidia fasciculata (CfTpx), and Salmonella typhimurium (AhpC) Prx orthologs using ClustalW. The conserved (identical or similar) amino acid residues are shown by shading; darker shading indicates identical residues, and lighter shading indicates similar residues. The location of the Cys residue corresponding to the Cys 83 of human Prx1 is marked with an asterisk. JULY 27, 2007 • VOLUME 282 • NUMBER 30 evidence to support the existence of inherent structural and functional differences between Prx1 and Prx2 at the protein level. We show that the differences between the two highly homologous proteins are attributable in part to the unique presence of Cys 83 in Prx1. The significance of this study is underlined by the realization that the differential molecular characteristics of Prx1 and Prx2 would continue to influence their molecular behaviors in various biological systems and also under a diverse environmental and genetic context. The information obtained in this study would provide a conceptual framework for further studies to delineate the physiological (or pathophysiological) functions of Prx1 and Prx2.